31 template <
class ArcT = fst::LexicographicArc< fst::StdArc::Weight, fst::StdArc::Weight> >
    44     stats ( new StatsData ),
    54   unordered_map<std::size_t, std::string> 
oovwmap;
    67   unordered_map<std::string, std::vector< pair <unsigned, unsigned> > > 
hpinstances;
    73   unordered_set<std::string> 
tvcb;
    80   std::vector< fst::VectorFst<ArcT> *> 
filters;
    84   unordered_map<std::string, void * > 
fsts;
    86   inline fst::VectorFst<ArcT> *
getFst(std::string 
const &key) {
    87     unordered_map<std::string, void * >::iterator itx
    89     if ( itx == fsts.end() ) {
    90       LWARN(
"Empty lattice key=" << key);
    93     return reinterpret_cast<fst::VectorFst<ArcT> *
>(itx->second);
    99   unordered_map<std::string, std::vector <const KenLMData *> > 
klm;
   116   unordered_map<std::string, WordMapper *> 
wm;
 
unordered_map< std::size_t, std::string > oovwmap
Contains oovs. 
Data structure containing all cyk-related information. 
unordered_map< uint, std::string > grammar_inversecategories_t
std::vector< fst::VectorFst< ArcT > * > filters
unordered_map< std::string, std::vector< const KenLMData * > > klm
Collections of language models accessed by keys (e.g. in translation we need a bunch for hifst and on...
boost::shared_ptr< ucam::fsttools::StatsData > stats
To collect statistics across the whole pipeline. 
Struct containing grammar rules. 
unordered_set< std::string > * recasingvcblm
mixed-case vocabulary of the recasing unigram language model 
unordered_map< std::string, std::vector< pair< unsigned, unsigned > > > hpinstances
std::string * translation
Translated sentence will be stored here. 
std::string originalsentence
source sentence 
unordered_map< std::string, WordMapper * > wm
Wordmap/Integer map objects. 
unsigned sidx
Sentence index. 
boost::shared_ptr< StatsData > stats
To collect statistics across the whole pipeline. 
const GrammarData * grammar
Contains translation grammar. 
std::string tokenizedsentence
grammar_inversecategories_t vcat
This information used for stats. 
Loads efficiently a wordmap file and provides methods to map word-to-integer or integer-to-word. To avoid memory footprint issues, hashing the wordmap entries is avoided. 
unordered_map< std::string, void * > fsts
Pointers to lattices (e.g. translation lattice, lmbr, etc) , and related, accessed by unique keys...
std::vector< std::string > pinstances
Pattern instances. 
unsigned numlocallm
Number of local language models used in hifst. 
Structure for sentence-specific grammar Rules will be queried by cyk per position and number of eleme...
unordered_set< std::string > tvcb
Target vocabulary. 
SentenceSpecificGrammarData * ssgd
Sentence-specific grammar information – hashes to rule indices. 
fst::VectorFst< ArcT > * getFst(std::string const &key)
CYKdata * cykdata
cyk data structures